Edge detection using generalized higher-order statistics

Abstract

A local operator is proposed which is able to extract the edges in an image through the evaluation of generalized higher-order statistical moments of the data. These moments are used for analyzing the asymmetry of the distribution of the data present in a small mask which scans the image. The advantage of the proposed algorithm is its robustness with respect to symmetrically distributed noise. Experimental results are reported which confirm the validity of the approach.<>